Content uploaded by Roger E Beaty
Author content
All content in this area was uploaded by Roger E Beaty on Apr 18, 2019
Content may be subject to copyright.
1
© 2018, American Psychological Association. This paper is not the copy of record and may not
exactly replicate the final, authoritative version of the article. Please do not copy or cite without
authors' permission. The final article will be available, upon publication, via its DOI:
10.1037/aca0000215
Creativity Assessment in Neuroscience Research
Mathias Benedek1,*, Alexander P. Christensen2, Andreas Fink1, & Roger E. Beaty3
1 Institute of Psychology, University of Graz, Austria; BioTechMed Graz
2 Department of Psychology, University of North Carolina at Greensboro, USA
3 Department of Psychology, Penn State University, USA
Accepted in Psychology of Aesthetics, Creativity, and the Arts
Special Issue: “Creativity Assessment - Pitfalls, Solutions, and Standards”
*Corresponding author:
Dr. Mathias Benedek, Institute of Psychology, University of Graz. Universitätsplatz 2, 8010 Graz,
Austria. Phone: +43 316 380 8475. E-mail: mathias.benedek@uni-graz.at.
Cite as: Benedek, M., Christensen, A. P., Fink, A., & Beaty, R. E. (in press). Creativity assessment in
neuroscience research. Psychology of Aesthetics, Creativity, and the Arts. Advance online publication.
doi: 10.1037/aca0000215
2
Abstract
The investigation of the neural correlates of creative cognition requires researchers to adapt
creativity tasks to meet the constraints imposed by cognitive neuroscience research – assessing well-
defined cognitive processes, repeated over many tasks. We present a brief review of essential study
design parameters in neuroscience research on creativity including number of task repetitions (i.e.,
trials), time on task, what kind of responses are collected (e.g., whether participants speak, write,
draw, press buttons etc.), and when these responses are collected (e.g., after or during task). We
further examine how design parameters depend on neuroscience methods (e.g., fMRI, EEG) and task
type (e.g., divergent thinking, creative problem solving). The review discloses a substantial
heterogeneity of methodological approaches across studies, but also identifies some established
common practices. Typical adaptations include the employment of shortened tasks, which allows the
realization of more tasks per session, and a more focused investigation of time-critical cognitive
processes. Study designs also commonly separate periods of creative thought from response
production in order to restrict the effect of response-related motor artifacts and to assess brain
activity unique to the generation of creative ideas/solutions. We discuss the pros and cons of the
various approaches with respect to the goal to increase reliability of neurophysiological
measurements while maintaining valid assessments, and derive some recommendations for future
research.
Keywords: Assessment; Creativity; Creative Cognition; Brain; EEG; fMRI
3
Neuroscience provides powerful methods to investigate cognitive processes in creative
thought. In the last two decades, many scholars have joined the venture to unveil how creativity
emerges in the brain. A special issue in the journal Methods supported these efforts by introducing
novel approaches to the study neuroscience of creativity (Dietrich, 2007). Early reviews of research
findings, however, observed little consistency in how creativity is manifested in the brain, which was
partly attributed to the large heterogeneity of tasks and methods employed in the field (Arden et al.,
2010; Dietrich et al., 2010; Sawyer, 2011). Indeed, studies have investigated very diverse creative
activities ranging from drawing to musical improvisation to idea generation (Abraham, 2018).
Moreover, even studies focusing on the same experimental task (e.g. alternate uses test) differed
considerably in the specific way the task was implemented in the neurophysiological assessment
(e.g., task duration, mode of responding, control tasks, etc.). Later reviews that put a focus on more
specific creative activities and brain imaging methods provided more consistent results (e.g., Boccia
et al., 2015; Fink & Benedek, 2014; Gonen-Yaacovi et al., 2013; see also, Jung & Vartanian, 2018).
However, experimental designs still vary considerably across studies even within specific domains,
which may reflect a lack of clarity of how to assess creativity most effectively in neuroscience
research.
The valid assessment of creativity is already a big challenge in behavioral creativity research.
All the more, this applies to cognitive neuroscience, because neurophysiological assessments impose
additional methodological constraints (Abraham, 2013; Sawyer, 2011). Brain research requires the
measurement of brain activation during well-defined cognitive activities that need to be repeatable
over dozens of times in order to obtain reliable assessments. Moreover, neurophysiological
assessments are sensitive to motor artifacts (e.g., caused by speaking, moving, or even blinking),
which seriously limits the feasible ways of creative expression. Finally, neuroscientific investigations
often involve test settings that appear not very conducive to creative thought (e.g. lying supine in a
rattling MRI scanner or being wired with EEG electrodes). This general situation is quite at odds with
the assessment of creative cognition, which typically represents complex cognitive activities that can
4
take minutes and often require the generation of elaborated ideas and products. Therefore,
neuroscience research on creativity is challenged to adapt creativity assessments to meet the
constraints imposed by neurophysiological measurements. This article provides a brief review of how
methodological issues in this field have been addressed so far. It focuses on functional imaging
studies that investigate brain activation during actual creative task performance. The article hence
does not cover other important lines of neuroscience research such as structural imaging, resting-
state analyses, or lesion studies that measure creativity independent from brain assessments. These
latter approaches face the same psychometric challenges as behavioral research on creativity, which
are addressed by other articles in this Special Issue. We discuss the pros and cons associated with
different methodological approaches, and derive some recommendations on how creativity can be
studied in a reliable and valid way in cognitive neuroscience research.
Reviewing common practices in neuroscience research on creativity
General overview
To obtain a comprehensive picture of how neuroscience studies have approached creativity
so far, we performed a quantitative review of available research in this field. We were interested in
studies that investigated brain activity during creative cognition using common neuroscience
methods including functional magnetic resonance imaging (fMRI), electroencephalography (EEG),
near-infrared spectroscopy (NIRS), positron emission tomography (PET), and
magnetoencephalography (MEG). Specifically, we searched the literature using Web of Science™ for
English empirical articles employing a topic search considering title, abstract, and keywords with the
search string: TS = ((creativity OR "creative cognition" OR "creative thinking") AND (fMRI OR EEG OR
MEG OR NIRS OR PET)). Note that we used method-specific search cues instead of more general cues
such as “brain” or “neuroscience” because it resulted in a more focused search of empirical work.
This approach should provide a broad, representative overview of relevant work, but is not a fully
exhaustive search, as it may miss studies that used relevant tasks, but did not mention
5
“creativity/creative cognition/creative thinking” in the title, abstract or keywords. The search yielded
305 papers on October, 2nd, 2018. We excluded papers that included no original data (e.g., reviews),
did not involve creative thinking tasks (e.g., metaphor comprehension), or did not study brain activity
during actual creative task performance (e.g., structural MRI studies). This procedure left 115 articles
describing a total of 131 different studies or tasks (some articles had multiple studies or tasks),
which represented the final data considered in our analyses (data and scripts are provided at
https://osf.io/zfr7v/).
An analysis of publications per year shows that, besides the pioneering work by Colin
Martindale carried out in the 1970’s and 1980’s, neuroscience research on creativity didn’t gain
momentum until the end of 2000, with publication numbers increasing steadily ever since, and about
70% of articles having been published since 2010 (see Figure 1). From the studies included in this
review, 48.9% used EEG, 48.1% used fMRI, and two studies each used PET and NIRS, respectively.
While fMRI has only been used since 2005 in this field, it has outnumbered EEG in recent research
and strongly contributes to the overall trend of increasing neuroscience research on creativity. Only
one article combined two neuroimaging methods (i.e., EEG and fMRI; see Fink et al., 2009),
suggesting a need for more multimodal research. The sample sizes in these studies range from 7 to
250, with an average of 35.81 (SD = 36.41; median = 28.0).
In the next step, we examined the prevalence of different kinds of creativity assessments in
neuroscience research. We found that the majority of studies (51.1%) used divergent thinking (DT)
tasks, which require to generate creative ideas to open-ended problems (e.g., alternate uses task);
19.1% of studies employed creative problem solving (CPS) tasks, which have correct solutions and
often require a restructuring of the problem representation (e.g., Remote Associates Test, insight
tasks); 29.8% of studies used a variety of mostly product-based tasks, which assess creative
performances resulting in a creative product such as drawing, writing, or musical improvisation.
These rates are very similar to those obtained in a recent review of behavioral creativity research
based on a random stratified sample of 200 articles from 2009-2012 (Forgeard & Kaufman, 2016):
6
from 81 studies that directly assessed creative performance, 59.3% employed DT tasks (traditional or
complex), 9.9% employed CPS tasks, and 30.9% used other tasks rated with the consensual
assessment technique. This congruence (including a similar focus on DT tasks) suggests that the
popularity of these task types in the behavioral research tradition is mirrored in neuroscience
research.
We further analyzed the relative frequency of task domains (e.g., verbal, visual, music)
defined by the type of response in these tasks. This analysis revealed that most of the employed
creativity tasks collected verbal responses (73.3%), 17.6% were visual tasks (e.g., drawing), 7.6%
were musical tasks, and 1.5% used other or mixed modalities (e.g., freestyle rap). It needs to be
noted that the large set of verbal tasks was not homogenous. It included tasks that required verbal
creativity in a narrower sense, for example, metaphor generation or story writing, but also many
other tasks that just required a verbal response but without involving verbal creativity to the same
degree. The alternate uses task, for example, has been commonly labeled as a verbal task (Torrance,
1974), although it may not require much verbal creativity to find creative object uses (Benedek, Fink
& Neubauer, 2006). Only after an idea is generated, some verbal creativity may be involved in how
well an idea is sold (Forthmann et al., 2017), suggesting that even different modalities can be
prevalent at different stages of the task. In general, verbal responses are arguably the most
convenient way to communicate abstract ideas. We hence believe that the prevalence of verbal
creativity is overestimated in the literature, and this variability in task classification may contribute to
inconsistencies in research findings. Therefore, traditional approaches to classify task domains by the
modality of responses need to be reconsidered and shifted towards a focus on the modality of
cognitive representations during the actual task.
How to adapt creativity tasks in neuroscience research?
Number of tasks and time on task. Neuroscience scholars commonly seek to employ proved and
tested tasks similar to how they are used in behavioral research, but several adaptions are needed to
7
meet the requirements of neurophysiological measurements. First of all, tasks are commonly
embedded in more extended trials that can include task cues (i.e., reminding participants of the
current task and condition, since thorough task instructions are often administered before entering
the scanner or mounting EEG electrodes), reference periods (i.e., measurements of baseline brain
activity, which often ask participants to relax while looking at a fixation cross), the actual task, and a
response period. Moreover, neuroscience assessments typically involve many task repetitions (i.e.,
trials) and shorter task durations than behavioral assessments. The main reason for this is that
neuroscience assessments do not target the measurement of a latent ability but rather the
assessment of brain processes related to creative task performance. Establishing the relationship
between creative cognition and brain activity involves several assumptions that are sometimes not
well specified. First of all, we need to define what cognitive processes are involved when in the
creative task. Yet, creative thinking tasks are often quite complex and involve many different
processes and strategies that may vary over time (e.g., Gilhooly, Fioratou, Anthony & Wynn, 2007).
Second, we need to define how cognitive processes manifest in a neurophysiological response.
Different neurophysiological responses are commonly considered, including event-related potentials
(ERP), and shifts in oscillatory activity in the EEG, or blood-oxygen-level dependent (BOLD) responses
in fMRI assessments, and different responses follow different temporal dynamics. Crucially, neural
responses of cognitive processes cannot be observed in isolation, but other ongoing processes as
well as measurement artifacts (e.g. motion-related artifacts) add noise to the signal we are
interested in. Further noise is contributed by inter- and intraindividual differences in cognitive
processes and brain responses, which are typically hard to account for (Grady & Garrett, 2014). Given
these potential sources of measurement error, many task repetitions (i.e., trials) are needed to
ensure sufficient reliability of neurophysiological assessments. A large number of trials can only be
realized with short tasks, which implies considerable shortening for extended creativity tasks (e.g.
divergent thinking tasks). Shorter tasks may additionally reduce the variability of cognitive processes
within tasks and better conform to neurophysiological models of brain responses.
8
Table 1 presents a descriptive analysis of the number of tasks and task duration (defined as
the time needed to generate a solution or product, which excludes any time periods designated for
instruction or response elaboration) in relevant research. Across all studies, the median number of
tasks (i.e., trials) was 15 and the median task duration was 30 s. Notably, the correlation between
number and duration of tasks showed a strong negative correlation (rs = -.75), indicating that studies
using shorter tasks afforded substantially more tasks. Task number and duration vary substantially
across studies and differ between imaging methods and task type. For example, fMRI studies
generally used more tasks (median = 20) than EEG studies (median = 3), but task duration was
shorter in fMRI studies (median = 15 s) compared to EEG studies (median = 165 s). In part, this
difference may be due to the fact that EEG assessments involve a less obtrusive assessment setting,
so that tasks can be implemented more akin to standard cognitive testing. Moreover, EEG studies
often target at the analysis of changes in brain activation over time within tasks (e.g., Jung-Beeman
et al., 2004; Schwab, Benedek, Papousek, Weiss, & Fink, 2012). Studies on creative problem solving
have used the highest number of tasks, followed by divergent thinking and other tasks, whereas time
on task was typically higher for divergent thinking tasks compared to creative problem solving. These
findings reflect differences in task duration that are also observed in behavioral testing (e.g., creative
problem solving tasks are often realized with shorter time on task than divergent thinking tasks).
Some studies even used as few as one task, which may have been done to establish basic brain
activity or connectivity patterns for a complex task (e.g., De Pisapia, Bacci, Parrott, & Melcher, 2016).
Some of the task durations are also very long (from 30 s to several minutes), which typically reflects
self-paced tasks that require ongoing production such as drawing, musical improvisation, or tasks
where multiple responses can be given within the task, which were later split into separate trials.
Our analysis shows that neuroscience studies on creativity typically employ around 15 tasks
(i.e., trials) per condition or task type. This is much more than the number of tasks commonly
employed in psychometric testing of the same constructs, but probably still less than the average
number of tasks used in other fields of cognitive neuroscience (Abraham, 2013). The adequate
9
number of tasks likely depends on the signal-to-noise ratio of the task and further design
characteristics such as task duration and event-related vs. blocked task presentation (Friston, 1999).
Moreover, for fMRI research, power calculation tools are available to determine the required sample
size for a given design (Mumford & Nichols, 2008). High task numbers are only possible with short
tasks, which comes with certain limitations. For example, divergent thinking tasks probably cannot be
much shorter than 10-15 s to enable at least one valid response. Task periods of 15 s easily add up to
a trial duration of 45 s, including time for cues, response periods, and fixation periods. If we have just
two conditions with 20 trials each, we already end up with a test session of 30 minutes. In fMRI
studies, we often want to add short structural scans and include time for initial calibration, which
easily reaches the maximum time that participants can stay engaged in a task and be exposed to MRI
measurements (i.e., usually about 45 minutes). Tasks focusing on more elementary cognitive aspects
underlying creative thought, like passive conceptual expansion during evaluation (Rutter et al., 2012)
or association processes (Green et al., 2015), allow shorter task times and thus higher amounts of
trials. Hence, increasing the number of trials is limited by the minimum amount of time needed to
perform tasks and the maximum total session time.
Besides these technical restrictions to tweaking tasks for neuroscience research, there are
also important conceptual issues to consider. How does the shortening of tasks affect task validity? In
DT tasks, for example, creativity of responses tend to increase with time as early responses are often
retrieved from memory and thus represent more common, less creative ideas (Beaty & Silvia, 2012;
Gilhooly et al., 2007). Yet, there is evidence suggesting that, while reliability increases with time on
task and number of responses, this may not equally apply to validity; in fact, task validity may even
decrease when DT tasks get too long (Benedek, Mühlmann, Jauk, & Neubauer, 2013). Moreover, we
still know little about how cognitive task demands change when tasks become more speeded (i.e.,
shorter task durations). One recent study showed that more speeded tasks (e.g., 2 minutes of
divergent thinking) did not rely more on mental speed (Gs) than “unspeeded” tasks (e.g., 8 minutes
of divergent thinking; Forthmann, Lips, Szardenings, Scharfen, & Holling, in press), but task durations
10
can be considerably shorter than that in neuroscience research. A few studies have reported
relationships between task performance during short tasks used in neuroscience assessments and
external criteria. For example, Perchtold and colleagues (2018) found a high correlation between
rated divergent thinking performance in 15 s tasks administered inside the scanner and 3 min tasks
administered outside the scanner (r = .64). Similarly, unpublished latent variable analyses from Beaty
and colleagues (2018) found a strong latent correlation of r = .61 between divergent thinking ability
assessed with 12 s scanner tasks (people generated 23 single uses for 23 different objects) and 3 min
lab tasks (people generated many creative uses for two objects), which again strongly correlated with
self-reported creative behavior and accomplishments in the arts and sciences (scanner r = .50 and lab
r = .32), providing validity evidence consistent with, and potentially even higher than, lab-based
measures (e.g., Jauk, Benedek & Neubauer 2014).
Collecting responses. Another aspect that requires careful consideration when adapting creativity
tasks to neuroscientific assessments is how and when responses are collected. Since
neurophysiological measurements are sensitive to movement artifacts, measurements of brain
activity should not get confounded with response-related motor activity. While subtle motor activity
related to button presses is typically not considered a problem, creative products often require more
complex response forms such as speaking, writing, drawing, or even playing an instrument. Artifact
detection and correction tools do an increasingly good job to remove motor-related artifacts from
the data, yet there are technical limits to restoring original data. Therefore, researchers typically
ensure to separate creative thought from response production in time. This approach is also
consistent with the aim to distinguish different phases in the creative process (e.g., generation,
elaboration, evaluation; Barbot, 2018; Ellamil et al., 2012; Fink et al., 2018; Jankowska et al., 2018;
Loesche et al., 2018; Rominger et al., 2018).
There are different ways to deal with this issue. First, we can have participants perform the
task self-paced, thus allowing them to give their responses whenever they arise, and only afterwards
individually classify times prior to the response as creative thinking time versus times after response
11
onset as response production time. Adapted to neuroscientific assessments this could mean that
participants press a button when they have an idea, then vocalize their idea, and press another
button when they are done with the response, thereby providing time stamps for thinking and
response periods (e.g., Boot et al., 2017; Fink, Benedek, Grabner, Staudt, & Neubauer, 2007). After a
response, they may move on to the next task, or, if multiple responses are required, continue with
the task until timeout. Another option is to use voice key analyses applied to audio recordings during
task performance to obtain relevant timings (e.g., Benedek et al., 2014b). Yet another approach is to
define fixed time intervals a priori for idea generation and response generation, respectively (e.g.,
Fink et al., 2009). Some researchers additionally ask participants to indicate the occurrence of
solutions by button-presses during task performance with fixed timing (e.g., Heinonen et al., 2018).
Finally, for some complex creative activities, response production cannot be easily delayed and brain
activation is measured concurrent to the actual production of responses (e.g. musical improvisation
or drawing; e.g., Pinho et al., 2016; Saggar et al., 2018). Such studies typically employ control tasks
that involve highly similar response productions in order to limit the risk of systematic confounds by
motor activity. Another question relates to when and how responses are collected. Some studies
collected verbal (Camarda et al., 2018) or written responses (e.g., Erhard, Kessler, Neumann, Ortheil
& Lotze, 2014), while others ask participants to draw (e.g., Saggar et al., 2018), or play piano using
MR-compatible instruments (Pinho et al., 2016). Yet other studies ask for button presses to indicate
that a solution was found, which is often followed up by asking participants to recall their responses
after the session or by assessing independent performance measures (e.g., Abraham et al., 2012;
Vartanian et al., 2018).
Table 2 presents an analysis of the typical response timing (i.e., when are responses
collected) and response type (i.e., what type of responses are collected) across neuroscience studies
on creativity. Most commonly, studies have defined a fixed timing for thinking and response periods,
either without button presses (40.1%) or with additional button presses during task performance to
measure response times (16.3%). This method is especially popular in fMRI studies (73.0% in total). It
12
is common across all types of tasks, but creative problem solving studies have used fixed timing more
often together with additional button press (52.5%) and other tasks more often without. 29.3% of
studies employed a self-paced approach, where the separation between thinking and response
periods is realized post hoc relative to individual response onsets. This approach is especially
common in EEG research (47.4%). Only 12.2% of studies have collected responses during the task,
thus not separating thinking periods from response periods.
The self-paced approach aims to implement tasks as similar as possible to how they are
typically administered in psychometric tests, allowing participants to respond whenever they come
to a solution (Fink et al., 2007). We can assume that this approach achieves a validity similar to
performance in standard cognitive testing. As a potential downside, this approach results in a
variable amount of time spent on the task across participants (if the task ends as soon as a response
is given; e.g., Tik et al., in press) or in a variable number of trials (if participants can give several
responses within each task; e.g., Boot et al., 2017). When a variable amount of data is collected per
participant this entails differences in the reliability of assessments, and it may even imply a direct
confound, as more creative people typically respond more fluently. Following this approach,
experimenters should make sure to obtain a certain minimum amount of trials/time for each
participant.
The fixed-timing approach ensures that the number of task repetitions (i.e., trials) and time
spent per task is equal across participants, which implies higher experimental control and also
enables a more straightforward analysis of brain data. As a potential downside of this approach,
however, it is not easy to specify a timing that works well for all task conditions and participants – it
should be long enough to capture the essential process and provide all participants enough time to
come up with a response, yet it should not be too long, as this can result in idling when participants
find a response early in the task. The latter issue is commonly dealt with by asking participants to
continue searching for even more creative ideas or to continue adding details to their response until
the predefined time is over (e.g., Beaty, Silvia, Benedek, 2017; Benedek et al., 2014a).
13
Neuroscience studies also differ considerably in what kind of responses are collected. Most
commonly they relied on oral responses (46.4%), but sometimes they also collected written
responses or drawings, and about 25% of studies acquired button presses. Oral responses are most
frequently collected in EEG research (56.7%) but also common in fMRI studies (33.9%). The review
further shows that oral responses are most common in divergent thinking research (66.2%), whereas
creative problem solving research more often use button presses (58.3%). These rates are certainly
different from other fields of cognitive neuroscience, which mostly rely on button press responses.
These findings highlight that most studies make an effort to collect responses in a way that pays
tribute to the complex nature of creative productions—that is, creative performance cannot easily be
reduced to button presses and typically involves speaking, writing, drawing or other formats to
adequately communicate creative ideas and products.
Collecting responses in neuroscience of creativity often requires some extra effort (e.g.,
employing an MRI-compatible microphone), but is important for several reasons. First of all, it allows
the researcher to monitor task performance and thus, substantiate that participants have been
properly engaged in the task. Related to this, it facilitates the researcher’s ability to discard trials
from analysis where participants failed to follow instructions or to come up with a response. This is
common practice in cognitive neuroscience research, but only possible when responses were
collected in the first place. From the perspective of the participant, tasks may be more engaging
when they require a response, whereas doing tasks just mentally can be tedious and increase the risk
for mind wandering. As another important benefit, responses can be scored to obtain measures of
individual performance. Scoring creative performance typically involves evaluations by several raters
that show reasonable inter-rater-reliability. Performance data can serve to run manipulation checks
(e.g., how do experimental conditions affect creative performance?; Benedek et al., 2018; Fink et al.,
2012), and they enable additional lines of analyses relating task performance to brain activation. This
can be done at the within-subjects level (i.e., how does brain activity differ between more vs. less
original ideas?; e.g., Green et al., 2015) as well as at the between-subjects level (i.e., how does brain
14
activity differ between more creative people vs. less creative people, as defined by task performance;
e.g., Fink & Neubauer, 2008). Finally, scored task performance can be related to relevant external
criteria in order to demonstrate the validity of adapted creativity tasks (Beaty et al., 2018). In sum,
assessing creative performance during neurophysiological assessments is challenging but crucial to
achieve a stronger inference on the actual relationship between creativity and brain activation.
Summary and conclusions
The investigation of the neural correlates of creative cognition requires researchers to adapt
creativity tasks to meet the constraints imposed by cognitive neuroscience research – assessing well-
defined cognitive processes, repeated over many tasks. A review of available neuroscience research
of creativity revealed a large variability in essential study design parameters such as number of tasks,
time on task, and when and what kind of responses are collected. In general, the employed creativity
tasks are strongly inspired by available psychometric tests of creative potential that were adapted
towards lower time on task to allow running more tasks per study. Most studies have collected
qualitative responses such as oral, written, drawn or musical productions, which enable to analyze
brain activation related to creative performance. Moreover, studies usually take care to separate
times of creative thought from response production in order to limit potential confounds with
response-related motor activity. Taken together these adaptations are useful to increase reliability of
neurophysiological measurements while maintaining valid assessments. Indeed, initial evidence
suggests that divergent thinking tasks that are adapted for neuroscience, for example, perform well
in terms of concurrent validity as well as criterion validity (Beaty et al., 2018; Perchtold et al., 2018).
While the development of neuroscience paradigms has strongly built on paradigms from
behavioral creativity research, we believe that neuroscience research also bears large potential to
inform behavioral research. First, neuroscience research requires great rigor regarding assumptions
on the involvement and timing of relevant cognitive processes. This has fueled new interest in the
examination of specific attention and memory processes in creative thought. For example, the robust
15
association between creativity and EEG alpha activity (Fink & Benedek, 2014; Kounios & Beeman,
2014) inspired cognitive research on internally directed attention in creative cognition (e.g., Ritter et
al., 2018; Salvi & Bowden, 2016; Walcher et al., 2017; for a review, see Benedek, 2018). Or, the
persistent relevance of the default network in creative cognition (Beaty et al., 2016; Zabelina &
Andrews-Hanna, 2016) attracted much interest in the role of episodic memory for creative thought
(e.g., Madore et al., 2016). Second, the constraints imposed by neuroscience research have
stimulated novel types and variants of creativity tasks. These tasks are often very well defined in
terms of their cognitive demands (Barbot, 2018; Prabhakaran, Green & Gray, 2014) and show
promising psychometric quality (e.g., Beaty et al., 2018). Hence, addressing the challenges for
assessment of creativity in neuroscience research has inspired creative solutions to creativity
assessment that may turn out to be more than just purposeful adaptations.
We conclude with some recommendations for future research. First, of course, it is crucial to
employ tasks that capture relevant aspects of creative cognition, thereby assuring optimal validity.
The demonstration of validity evidence (e.g., correlations with performance in original tasks or
external criteria of creativity) is essential for novel creativity tasks but is also recommended for
established tasks that have undergone substantial adaptations for neuroscience assessments.
Second, cognitive neuroscience research requires clear a priori assumptions on what cognitive
processes interact in creative tasks. Neuroscientific assessments of highly complex artistic
performances are intriguing, but they typically do not allow to reliably relate brain activation to
specific psychological processes and thus heavily rely on reverse inference (Poldrack, 2006). Powerful
tests of brain-cognition associations need to specify what cognitive processes (e.g., memory,
attention, cognitive control processes) are central to the main task, how they differ in control tasks
(but see Logothetis, 2008, for a discussion of pure insertion issues), and hypothesize on causal
relationships between neurocognitive processes (e.g., Vartanian et al., 2018). To this end, the design
of neurophysiological assessments needs to be guided by and based on the rich evidence of cognitive
science. Third, disentangling cognitive processes may further imply to distinguish specific phases or
16
stages in the creative process as presumed by available theoretical models (e.g., generation,
evaluation, elaboration). Fourth, creative cognition stands out in that it commonly involves the
generation of ideas or products that differ in quality. Assessing these productions and their creative
quality rather than just the time of their occurrence enables powerful analyses of creativity-related
brain functions. Fifth, co-registration studies assessing different neurophysiological parameters
concurrently are needed to relate and consolidate evidence across neuroscience methods. Finally, as
all fields of cognitive neurosciences, the neuroscience of creativity needs to ensure that studies are
well-powered in terms of sufficient trials and sample size (Yarkoni, 2009). In this manner, task-based
creativity neuroscience, together with other techniques including structural imaging (Jung et al.,
2013), brain stimulation (Weinberger et al., 2017), and neuropsychological approaches (Abraham,
2018) will help us to advance our understanding of how creativity emerges in the brain.
17
References
Abraham, A. (2018). The neuropsychology of creativity. Current Opinions in Behavioral Sciences, 27,
71-76.
Abraham, A. (2018). The neuroscience of creativity. Cambridge: Cambridge University Press.
Abraham, A. (2013). The promises and perils of the neuroscience of creativity. Frontiers in Human
Neuroscience, 7, 246.
Abraham, A., Pieritz, K., Thybusch, K., Rutter, B., Kröger, S., Schweckendiek, J., ... & Hermann, C.
(2012). Creativity and the brain: uncovering the neural signature of conceptual expansion.
Neuropsychologia, 50, 1906-1917.
Arden, R., Chavez, R. S., Grazioplene, R., & Jung, R. E. (2010). Neuroimaging creativity: a psychometric
view. Behavioural Brain Research, 214, 143-156.
Barbot, B. (2018). Measuring the dynamics of creative ideation: A new assessment paradigm.
Presentation at the Meeting of the Society for the Neuroscience of Creativity. Cambridge,
MA.
Beaty, R.E., Benedek, M., Silvia, P.J., & Schacter, D.L. (2016). Creative cognition and brain network
dynamics. Trends in Cognitive Sciences, 20, 87-95.
Beaty, R.E., Kenett, Y.N., Christensen, A.P., Rosenberg, M.D., Benedek, M., Chen, Q., Fink, A., Qiu, J.,
Kwapil, T., Kane, M.J., & Silvia, P.J. (2018). Robust prediction of individual creative ability
from brain functional connectivity. Proceedings of the National Academy of Sciences (PNAS),
115, 1087-1092.
Beaty, R. E., & Silvia, P. J. (2012). Why do ideas get more creative across time? An executive
interpretation of the serial order effect in divergent thinking tasks. Psychology of Aesthetics,
Creativity, and the Arts, 6, 309-319.
18
Beaty, R.E., Silvia, P.J., & Benedek, M. (2017). Brain networks underlying novel metaphor production.
Brain and Cognition, 111, 163-170.
Benedek, M. (2018). Internally directed attention in creative cognition. In R. E. Jung & O. Vartanian
(Eds.), The Cambridge handbook of the neuroscience of creativity (pp. 180-194). New York:
Cambridge University Press.
Benedek, M., Beaty, R., Jauk, E., Koschutnig, K., Fink, A., Silvia, P. J., Dunst, B., & Neubauer, A.C.
(2014a). Creating metaphors: The neural basis of figurative language production.
NeuroImage, 90, 99-106.
Benedek, M., Fink, A. & Neubauer, A.C. (2006). Enhancement of ideational fluency by means of
computer-based training. Creativity Research Journal, 18, 317-328.
Benedek, M., Jauk, E., Fink, A., Koschutnig, K., Reishofer, G., Ebner, F., & Neubauer, A. C. (2014b). To
create or to recall? Neural mechanisms underlying the generation of creative new ideas.
NeuroImage, 88, 125-133.
Benedek, M., Mühlmann, C., Jauk, E., & Neubauer, A. C. (2013). Assessment of divergent thinking by
means of the subjective top-scoring method: Effects of the number of top-ideas and time-on-
task on reliability and validity. Psychology of Aesthetics, Creativity, and the Arts, 7, 341-349.
Benedek, M., Schües, T., Beaty, R., Jauk, E., Koschutnig, K., Fink, A., & Neubauer, A. C. (2018) To
create or to recall original ideas: Brain processes associated with the imagination of novel
object uses. Cortex, 99, 93-102.
Boot, N., Baas, M., Mühlfeld, E., de Dre, C.W.K., & van Gaal, S. (2017). Widespread neural oscillations
in the delta band dissociate rule convergence from rule divergence during creative idea
generation. Neuropsychologia, 104, 8-17.
19
Camarda, A., Salvia, E., Vidal, J., Weil, B., Poirel, N., Houde, O., ... & Cassotti, M. (2018). Neural basis
of functional fixedness during creative idea generation: an EEG study. Neuropsychologia, 188,
4-12.
De Pisapia, N., Bacci, F., Parrott, D., & Melcher, D. (2016). Brain networks for visual creativity: a
functional connectivity study of planning a visual artwork. Scientific reports, 6, 39185.
Dietrich, A. (Eds.) (2007). Neurocognitive Mechanisms of Creativity: A Toolkit [Special Issue].
Methods, 42(1).
Ellamil, M., Dobson, C., Beemna, M., & Christoff, K. (2012). Evaluative and generative modes of
thought during the creative process. NeuroImage, 59, 1783-1794.
Erhard, K., Kessler, F., Neumann, N., Ortheil, H. J., & Lotze, M. (2014). Professional training in creative
writing is associated with enhanced fronto-striatal activity in a literary text continuation task.
NeuroImage, 100, 15-23.
Fink, A., & Benedek, M. (2014). EEG alpha power and creative ideation. Neuroscience and
Biobehavioral Reviews, 44, 111-123.
Fink, A., Benedek, M., Grabner, R. H., Staudt, B. & Neubauer, A.C. (2007). Creativity meets
neuroscience: Experimental tasks for the neuroscientific study of creative thinking. Methods,
42, 67-75.
Fink, A., Grabner, R. H., Benedek, M., Reishofer, G., Hauswirth, V., Fally, M., Neuper, C., Ebner, F. &
Neubauer, A.C. (2009). The creative brain: Investigation of brain activity during creative
problem solving by means of EEG and fMRI. Human Brain Mapping, 30, 734-748.
Fink, A., Koschutnig, K., Benedek, M., Reishofer, G., Ischebeck, A., Weiss, E. M., & Ebner, F. (2012).
Stimulating creativity via exposure to other people’s ideas. Human Brain Mapping, 33, 2603-
2610.
20
Fink, A., & Neubauer, A. C. (2008). Eysenck meets Martindale: The relationship between extraversion
and originality from the neuroscientific perspective. Personality and Individual Differences,
44, 299-310.
Fink, A., Rominger, C., Benedek, M., Perchtold, C., Papousek, I., Weiss, E.M., Seidel, A., & Memmert,
D. (2018). EEG alpha activity during imagining creative moves in soccer decision-making
situations. Neuropsychologia, 114, 118-124.
Foregeard, M. J. C., & Kaufman, J. C. (2016). Who cares about imagination, creativity, and innovation
and why? A Review. Psychology of Aesthetics, Creativity, and the Arts, 10, 250-269.
Forthmann, B., Holling, H., Çelik, P., Storme, M., & Lubart, T. (2017). Typing speed as a confounding
variable and the measurement of quality in divergent thinking. Creativity Research Journal,
29, 257-269.
Forthmann, B., Lips, C., Szardenings, C., Scharfen, J., & Holling, H. (in press.) Are speedy brains
needed when divergent thinking is speeded - or unspeeded? The Journal of Creative
Behavior. https://doi.org/10.1002/jocb.350
Gilhooly, K.J., Fioratou, E., Anthony, S.H., Wynn, V. (2007). Divergent thinking: strategies and
executive involvement in generating novel uses for familiar objects. British Journal of
Psychology, 98, 611–625.
Grady, C. L., & Garrett, D. D. (2014). Understanding variability in the BOLD signal and why it matters
for aging. Brain Imaging and Behavior, 8, 274-283.
Green, A. E., Cohen, M. S., Raab, H. A., Yedibalian, C. G., & Gray, J. R. (2015). Frontopolar activity and
connectivity support dynamic conscious augmentation of creative state. Human brain
mapping, 36, 923-934.
21
Gonen-Yaacovi, G., De Souza, L. C., Levy, R., Urbanski, M., Josse, G., & Volle, E. (2013). Rostral and
caudal prefrontal contribution to creativity: a meta-analysis of functional imaging data.
Frontiers in Human Neuroscience, 7, 465.
Heinonen, J., Numminen, J., Hlushchuk, Y., Antell, H., Taatila, V., & Suomala, J. (2016) Default mode
and executive networks areas: Association with the serial order in divergent thinking. PLoS
ONE, 11: e0162234.
Jankowska, D. M., Czerwonka, M., Lebuda, I., & Karwowski, M. (2018). Exploring the creative process:
Integrating psychometric and eye-tracking approaches. Frontiers in Psychology, 9, 1931.
Jauk, E., Benedek, M., & Neubauer, A. C. (2014). The road to creative achievement: A latent variable
model of ability and personality predictors. European Journal of Personality, 28, 95-105.
Jung, R. E., Mead, B. S., Carrasco, J., & Flores, R. A. (2013). The structure of creative cognition in the
human brain. Frontiers in Human Neuroscience, 7, 330.
Jung, R. E. & Vartanian O. (Eds.). (2018). The Cambridge handbook of the neuroscience of creativity.
New York: Cambridge University Press.
Jung-Beeman, M., Bowden, E. M., Haberman, J., Frymiare, J. L., Arambel-Liu, S., Greenblatt, R., ... &
Kounios, J. (2004). Neural activity when people solve verbal problems with insight. PLoS
Biology, 2, e97.
Kounios, J., & Beeman, M. (2014). The cognitive neuroscience of insight. Annual Review of
Psychology, 65, 71-93.
Loesche, F., Goslin, J., & Bugmann, G. (2018). Paving the Way to Eureka—Introducing “Dira” as an
Experimental Paradigm to Observe the Process of Creative Problem Solving. Frontiers in
Psychology, 9, 1773.
Logothetis, N. K. (2008). What we can do and what we cannot do with fMRI. Nature, 453, 869-878.
22
Madore, K. P., Jin, H., G., & Schacter, D. L., (2016). Divergent creative thinking in young and older
adults: Extending the effects of an episodic specificity induction. Memory & Cognition, 44,
974-988.
Perchtold, C. M., Papousek, I., Koschutnig, K., Rominger, C., Weber, H., Weiss, E. M., & Fink, A.
(2018). Affective creativity meets classic creativity in the scanner. Human Brain Mapping, 39,
393-406.
Poldrack, R. A. (2006) Can cognitive processes be inferred from neuroimaging data? Trends in
Cognitive Sciences, 10, 59-63.
Prabhakaran, R., Green, A. E., & Gray, J. R. (2014). Thin slices of creativity: Using single-word
utterances to assess creative cognition. Behavior Research Methods, 46, 641-659.
Rominger, C., Papousek, I., Perchtold, C.M., Weber, B., Weiss, E.M., & Fink, A. (2018). The creative
brain in the figural domain: Distinct patterns of EEG alpha power during idea generation and
idea elaboration. Neuropsychologia, 118, 13-19.
Ritter, S.M., Abbing, J. & van Schie, H.T. (2018) Eye-closure enhances creative performance on
divergent and convergent creativity tasks. Frontiers in Psychology, 9, 1315.
Rutter, B., Kröger, S., Hill, H., Windmann, S., Hermann, C., & Abraham, A. (2012). Can clouds dance?
Part 2: An ERP investigation of passive conceptual expansion. Brain and Cognition, 80, 301-
210.
Saggar, M., Quintin, E. M., Kienitz, E., Bott, N. T., Sun, Z., Hong, W. C., ... & Hawthorne, G. (2015).
Pictionary-based fMRI paradigm to study the neural correlates of spontaneous improvisation
and figural creativity. Scientific Reports, 5, 10894.
Salvi, C., & Bowden, E. M. (2016). Looking for creativity: Where do we look when we look for new
ideas? Frontiers in Psychology, 7, 161.
23
Sawyer, K. (2011). The cognitive neuroscience of creativity: a critical review. Creativity Research
Journal, 23, 137-154.
Schwab, D., Benedek, M., Papousek, I., Weiss, E.M., & Fink, A. (2014). The time-course of EEG alpha
power changes in creative ideation. Frontiers in Human Neuroscience, 8, 310.
Tik, M., Sladky, R., Luft, C. D. B., Willinger, D., Hoffmann, A., Banissy, M. J., ... & Windischberger, C. (in
press). Ultra‐high‐field fMRI insights on insight: Neural correlates of the Aha!‐moment.
Human Brain Mapping.Torrance, E. P. (1974). Torrance Tests of Creative Thinking: Norms,
technical manual, verbal forms A and B. Bensenville, IL: Scholastic Testing Service.
Vartanian, O., Beatty, E. L., Smith, I., Blackler, K., Lam, Q., & Forbes, S. (2018). One-way traffic: The
inferior frontal gyrus controls brain activation in the middle temporal gyrus and inferior
parietal lobule during divergent thinking. Neuropsychologia, 118, 68-78..Walcher, S., Körner,
C., & Benedek, M. (2017). Looking for ideas: Eye behavior during goal-directed internally-
focused cognition. Consciousness and Cognition, 53, 165-175.
Weinberger, A. B., Green, A. E., & Chrysikou, E. G. (2017). Using transcranial direct current
stimulation to enhance creative cognition: interactions between task, polarity, and
stimulation site. Frontiers in Human Neuroscience, 11, 246.
Yarkoni, T. (2009). Big correlations in little studies: Inflated fMRI correlations reflect low statistical
power—Commentary on Vul et al. (2009). Perspectives on Psychological Science, 4, 294-298.
Zabelina, D. L., & Andrews-Hanna, J. R. (2016). Dynamic network interactions supporting internally-
oriented cognition. Current Opinion in Neurobiology, 40, 86-93.
24
TABLES
Table 1
Analysis of the number of tasks (i.e., trials) and task duration [s] in neuroscience studies on creativity
(1975-2018). Descriptive statistics for all studies, and separate for EEG and fMRI studies, and studies
on divergent thinking (DT), creative problem solving (CPS), or other tasks (i.e., mostly product-based
tasks such as creative writing, drawing or musical improvisation).
# Tasks
n
Median
M
SD
Min
Max
All studies
117
15
24.53
33.74
1
210
EEG
53
3
12.62
26.07
1
150
fMRI
61
20
35.87
36.54
1
210
DT
60
13.5
15.63
14.76
1
72
CPS
22
33
42.82
40.65
1
150
Other tasks
35
10
28.29
45.98
1
210
Time on task [s]
n
Median
M
SD
Min
Max
All studies
117
30
132.53
434.91
2
4500
EEG
54
165
242.91
616.95
4
4500
fMRI
60
15
36.81
95.89
2
600
DT
62
30
154.24
576.27
2
4500
CPS
20
19
140.18
231.00
4.5
900
Other tasks
35
30
89.69
130.02
2
480
Note. n represents the number of valid data in each analysis, after excluding studies that failed to
report relevant information.
25
Table 2
Analysis of when and what kind of responses are collected in neuroscience studies on creativity (1975-
2018). Descriptive statistics for all studies, and separate for EEG and fMRI studies, and studies on
divergent thinking (DT), creative problem solving (CPS), or other tasks.
Response timing
[%]
n
After task
(Self-paced)
After task
(Fixed)
After task
(Fixed + RT)
During
task
No
response
All studies
123
29.3
40.1
16.3
12.2
1.6
EEG
57
47.4
33.3
5.3
12.3
1.8
fMRI
63
12.7
46.0
27.0
12.7
1.6
DT
64
35.9
48.4
6.3
7.8
1.6
CPS
23
34.8
13.0
52.2
0.0
0.0
Other tasks
36
13.9
44.4
11.1
27.8
2.8
Response type
[%]
Speak
Write
Draw
Button
Other
None
All studies
125
46.4
12.0
12.8
23.2
4.8
0.8
EEG
60
56.7
18.3
16.7
5.0
1.7
1.7
fMRI
62
33.9
6.5
9.7
41.9
8.1
0.0
DT
65
66.2
13.8
7.7
12.3
0.0
0.0
CPS
24
33.3
8.3
0.0
58.3
0.0
0.0
Other tasks
36
19.4
11.1
30.6
19.4
16.7
2.8
Notes. n represents the number of valid data in each analysis, after excluding studies that failed to
report relevant information. After task (self-paced) = participants can give their response at any time,
with the response onset marking the end of generation period and the beginning of the response
period; after task (fixed) = fixed thinking time and response time were specified by the experimenter;
after task (fixed + RT) = fixed thinking time and response time were specified by the experimenter,
but participant gives button press when a solution occurs; during task = responses were collected
concurrently to task performance.
26
Figure 1. Top: Number of articles on the neuroscience of creativity over time per method (dark blue =
EEG, blue = fMRI, light blue = other method). *Note that time bins represent 5 years, except for the
last, most recent bin, which only covers a little more than 3 years, i.e., 01/2015-10/2018. Bottom left:
Relative frequency of neuroscience studies employing fMRI, EEG, or other methods. Bottom right:
Relative frequency of neuroscience studies investigating divergent thinking, creative problem solving,
or other tasks.
A preview of this full-text is provided by American Psychological Association.
Content available from Psychology of Aesthetics Creativity and the Arts
This content is subject to copyright. Terms and conditions apply.